CI build failure in v6.6-rt

CI build failure in v6.6-rt

[Clark Williams]

Greg,

Kernel CI is reporting a build failure on v6.6-rt:

https://grafana.kernelci.org/d/build/build?orgId=1&var-datasource=default&var-build_architecture=riscv&var-build_config_name=defconfig&var-id=maestro:66a6b448bb1dfd36a925ebef

It's in arch/riscv/kernel/cpufeature.c where a return statement in
check_unaligned_access() doesn't have a value (and
check_unaligned_access returns int).

Is 6.6 stable supporting RiscV? If so then we either have to fix that
return, or backport the refactor of arch/riscv/kernel/cpufeature.c
(f413aae96cda0).  If it's not then who should I talk to about turning
off riscv CI builds for v6.6-rt?

Thanks,
Clark

Re: CI build failure in v6.6-rt

[Kroah-Hartman, Greg]

On Fri, Aug 02, 2024 at 08:46:55PM +0000, Clark Williams wrote:
> Greg,
> 
> Kernel CI is reporting a build failure on v6.6-rt:
> 
> https://grafana.kernelci.org/d/build/build?orgId=1&var-datasource=default&var-build_architecture=riscv&var-build_config_name=defconfig&var-id=maestro:66a6b448bb1dfd36a925ebef
> 
> It's in arch/riscv/kernel/cpufeature.c where a return statement in
> check_unaligned_access() doesn't have a value (and
> check_unaligned_access returns int).
> 
> Is 6.6 stable supporting RiscV? If so then we either have to fix that
> return, or backport the refactor of arch/riscv/kernel/cpufeature.c
> (f413aae96cda0).  If it's not then who should I talk to about turning
> off riscv CI builds for v6.6-rt?

Why not ask the people responsible for the -rt patchset?  If this isn't
an issue on a non-rt kernel, then I think you found the problem :)

thanks,

greg k-h

Re: CI build failure in v6.6-rt

[Sebastian Andrzej Siewior]

On 2024-08-03 08:20:16 [+0200], Kroah-Hartman, Greg wrote:
> On Fri, Aug 02, 2024 at 08:46:55PM +0000, Clark Williams wrote:
> > Greg,
> > 
> > Kernel CI is reporting a build failure on v6.6-rt:
> > 
> > https://grafana.kernelci.org/d/build/build?orgId=1&var-datasource=default&var-build_architecture=riscv&var-build_config_name=defconfig&var-id=maestro:66a6b448bb1dfd36a925ebef
> > 
> > It's in arch/riscv/kernel/cpufeature.c where a return statement in
> > check_unaligned_access() doesn't have a value (and
> > check_unaligned_access returns int).
> > 
> > Is 6.6 stable supporting RiscV? If so then we either have to fix that
> > return, or backport the refactor of arch/riscv/kernel/cpufeature.c
> > (f413aae96cda0).  If it's not then who should I talk to about turning
> > off riscv CI builds for v6.6-rt?
> 
> Why not ask the people responsible for the -rt patchset?  If this isn't
> an issue on a non-rt kernel, then I think you found the problem :)

I'm sorry. I misunderstood Clark and assumed this is a stable issue. Now
that I have the needed pointers, it is a RT-stable issue only.
I'm going to drop the stable folks from Cc: and reply with a patch.

> thanks,
> 
> greg k-h

Sebastian

Re: CI build failure in v6.6-rt

[Sebastian Andrzej Siewior]

On 2024-08-05 12:58:36 [+0200], To Kroah-Hartman, Greg wrote:
> I'm sorry. I misunderstood Clark and assumed this is a stable issue. Now
> that I have the needed pointers, it is a RT-stable issue only.
> I'm going to drop the stable folks from Cc: and reply with a patch.
> 

Clark, this one, please:

----->8------

From: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Date: Mon, 5 Aug 2024 12:38:59 +0200
Subject: [PATCH] riscv: Add return value to check_unaligned_access().

The stable backport of commit c20d36cc2a207 ("riscv: don't probe
unaligned access speed if already done") added a `return' stament to
check_unaligned_access(). The return value of the function is `void' but
the RT tree has the backport of commit 73cf6167ef44c ("RISC-V: Probe
misaligned access speed in parallel") which changes the return type to
`int'.

Make the return statement return 0.

Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
---
 arch/riscv/kernel/cpufeature.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/arch/riscv/kernel/cpufeature.c b/arch/riscv/kernel/cpufeature.c
index dd118773e717b..347cdf56f3a15 100644
--- a/arch/riscv/kernel/cpufeature.c
+++ b/arch/riscv/kernel/cpufeature.c
@@ -588,7 +588,7 @@ static int check_unaligned_access(void *param)
 
 	/* We are already set since the last check */
 	if (per_cpu(misaligned_access_speed, cpu) != RISCV_HWPROBE_MISALIGNED_UNKNOWN)
-		return;
+		return 0;
 
 	/* Make an unaligned destination buffer. */
 	dst = (void *)((unsigned long)page_address(page) | 0x1);
-- 
2.45.2

Sebastian

Re: CI build failure in v6.6-rt

[Clark Williams]

Thanks, I missed that we had changed the return value for
check_unaligned_access(). I'll apply this to the next stable update.

Clark

On Mon, Aug 5, 2024 at 11:00 AM Sebastian Andrzej Siewior
<bigeasy@linutronix.de> wrote:
>
> On 2024-08-05 12:58:36 [+0200], To Kroah-Hartman, Greg wrote:
> > I'm sorry. I misunderstood Clark and assumed this is a stable issue. Now
> > that I have the needed pointers, it is a RT-stable issue only.
> > I'm going to drop the stable folks from Cc: and reply with a patch.
> >
>
> Clark, this one, please:
>
> ----->8------
>
> From: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
> Date: Mon, 5 Aug 2024 12:38:59 +0200
> Subject: [PATCH] riscv: Add return value to check_unaligned_access().
>
> The stable backport of commit c20d36cc2a207 ("riscv: don't probe
> unaligned access speed if already done") added a `return' stament to
> check_unaligned_access(). The return value of the function is `void' but
> the RT tree has the backport of commit 73cf6167ef44c ("RISC-V: Probe
> misaligned access speed in parallel") which changes the return type to
> `int'.
>
> Make the return statement return 0.
>
> Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
> ---
>  arch/riscv/kernel/cpufeature.c | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/arch/riscv/kernel/cpufeature.c b/arch/riscv/kernel/cpufeature.c
> index dd118773e717b..347cdf56f3a15 100644
> --- a/arch/riscv/kernel/cpufeature.c
> +++ b/arch/riscv/kernel/cpufeature.c
> @@ -588,7 +588,7 @@ static int check_unaligned_access(void *param)
>
>         /* We are already set since the last check */
>         if (per_cpu(misaligned_access_speed, cpu) != RISCV_HWPROBE_MISALIGNED_UNKNOWN)
> -               return;
> +               return 0;
>
>         /* Make an unaligned destination buffer. */
>         dst = (void *)((unsigned long)page_address(page) | 0x1);
> --
> 2.45.2
>
> Sebastian
>